Is quality of experience beyond SLAs?

In this column I am going to press the point that quality of experience sets all traditional notions of service-level agreements on their heels. Now – not later – is the time to make the mental leap. QoE represents a fundamental shift in how SLAs can be defined.

Taken at face value, QoE is exactly what it sounds like – the quality of experience. “Experience” is defined in my Oxford American Dictionary as “an actual observation of facts and events,” and, as a verb, “to observe, to share, to actually be affected by – a feeling.” What’s interesting from these definitions is that the word itself combines two very different dimensions. One is a more empirical sense of observed reality, while the other includes sensation and imagination – it is about feeling.

Both definitions play in QoE – which reflects a very different agenda than traditional SLAs. Rather than simply building from what’s measurable up to the customer or end user, QoE would suggest starting with the end user, honoring the objective and subjective merits of his or her experience and trying to approximate them in metrics that can be validated in terms of technical performance and customer behavior.

You already may be thinking that this approach is an unhealthy combination of masochism and naiveté, but I would argue just the opposite – it is the shortest path to comfort and mental health for you and your customers.

Business productivity, customer loyalty, and business partnerships depend on QoE in all its dimensions. No one will stick with a provider that gets gold stars for SLAs but still leaves them experientially discontent – especially if other options present themselves. By trying to force you and your customers to live in a simulated universe in which only technical metrics apply, it is you who are being naïve. Sure, you will need to “manage” expectations and set some technical boundaries, but your ability to do this successfully is greatly enhanced once you approach the problem in terms of multi-dimensional experience rather than introverted technical specifications.

A few pointers and observations:

* Listen to your customers. While the old-fashioned help desk approach is often reactive and cumbersome, it can also provide useful background on customer perceptions and requirements. A strong, proactive service initiative will also help to promote dialog and interaction.

* Recognize that while availability and performance remain prime factors, there are other dimensions to QoE – such as consistency, cost to the customer, security, flexibility (e.g., mobility of a service, or customer choice of service), and variety (number of available and customer-relevant services). This is not a finite list – because the dimensions of experience are not finite.

* Look at options for testing responsiveness. Since degraded service has proven to be more of a customer turnoff than intermittent spurts of lack of availability, performance and QoE are probably the two most closely linked metrics. Until fairly recently, synthetic transaction analyses were the top choice for QoE validation, and they do still play a role. Synthetic transactions provide IT with a self-contained context for control. You can set the time and frequency and define SLAs accordingly – and of course synthetic transactions are superior for testing availability.

New technologies – including slimmer, more efficient agent technology, more advanced server-based transaction analysis, and significant advances in techniques for packet analysis – are making observed transactional baselining more possible. Unlike synthetic transactions, observed baselining can inform you, on a dynamic basis, of actual customer behavior and customer disaffection – for example, when transactions are aborted due to impatience. Some techniques are now highly scalable in capturing individual user behaviors as well as infrastructure performance in large, geographically dispersed environments.

These are just a few points. I invite your comments and opinions as well, and welcome intelligent disagreement and notes of support.

Oh, and to answer the question posed in the headline: in my opinion, the glorious and troublesome fact is that QoE is indeed beyond SLAs, which can only, at best, approximate it – and that’s because experience, itself, is more sprawling than the Internet, and more complex than all the data centers in the world.